Well let's see

ugropy/groupscsv/abdulelah_gani/primary/primary.csv

@@ -0,0 +1,221 @@
+group|smarts|molecular_weight|group_number
+CH3|[CX4H3]|15.035|1
+CH2|[CX4H2;!R]|14.027|2
+CH|[CX4H;!R]|13.019|3
+C|[CX4H0;!R]|12.011|4
+CH2=CH|[CH2]=[CH]|27.046|5
+CH=CH|[CH;!R]=[CH;!R]|26.038|6
+CH2=C|[CH2]=[CH0;!R]|26.038|7
+CH=C|[CH;!R]=[CH0;!R]|25.03|8
+C=C|[CH0;!R]=[CH0;!R]|24.022|9
+CH2=C=CH|[CH2]=[CH0;!R]=[CH;!R]|39.0564|10
+CH2=C=C|[CH2]=[CH0;!R]=[CH0;!R]|38.0486|11
+C=C=C|[CH0;!R]=[CH0;!R]=[CH0;!R]|36.033|12
+CH≡C|[CH]#[CH0]|25.0298|13
+C≡C|[CH0]#[CH0]|24.022|14
+aCH|[cH]|13.019|15
+aC fused with aromatic ring|[cH0;$([cH0]([a])([a])[a])]|12.011|16
+aC fused with non-aromatic ring|[cH0;$([cH0]([a])([a])[A;R])]|12.011|17
+aC except as above|[cH0;!$([cH0;$([cH0]([a])([a])[A;R])]);!$([cH0]([a])([a])[a]))]|12.011|18
+aN in aromatic ring|[n;R]|14.007|19
+aC-CH3|[cH0][CX4H3]|27.046|20
+aC-CH2|[cH0][CX4H2;!R]|26.038|21
+aC-CH|[cH0][CX4H;!R]|25.03|22
+aC-C|[cH0][CX4H0;!R]|24.0222|23
+aC-CH=CH2|[cH0][CH;!R]=[CH2]|39.0564|24
+aC-CH=CH|[cH0][CH;!R]=[CH;!R]|38.0486|25
+aC-C=CH2|[cH0][CH0;!R]=[CH2]|38.0486|26
+aC-C≡CH|[cH0][CH0;!R]#[CH]|37.0408|27
+aC-C≡C|[cH0][CH0;!R]#[CH0;!R]|36.033|28
+OH|[OH;$([OH][#6,Si])]|17.007|29
+aC-OH|[cH0][OH]|29.018|30
+COOH|[CH0](=O)[OH]|45.017|31
+aC-COOH|[cH0][CH0](=O)[OH]|57.028|32
+CH3CO|[CH3][CH0;$(C([CH3])-[#6]),$(C([CH3])-[Si])]=O|43.045|33
+CH2CO|[CH2][CH0;$(C([CH2])-[#6]),$(C([CH2])-[Si])]=O|42.037|34
+CHCO|[CH][CH0;$(C([CH])-[#6]),$(C([CH])-[Si])]=O|41.028841|35
+CCO|[CH0][CH0;$(C([CH0])-[#6]),$(C([CH0])-[Si])]=O|40.021|36
+aC-CO|[cH0][CH0;$(C([cH0])-[#6]),$(C([cH0])-[Si])]=O|40.021|37
+CHO|[CH;$([CH][#6,Si])](=O)|29.018|38
+aC-CHO|[cH0][CH](=O)|41.029|39
+CH3COO|[CH3;!R][C](=O)[OH0]|59.044|40
+CH2COO|[CH2;!R][C](=O)[OH0]|58.036|41
+CHCOO|[CH;!R][C](=O)[OH0]|57.0282|42
+CCOO|[CH0;!R][C](=O)[OH0]|56.0204|43
+HCOO|[CH](=O)[OH0]|45.017|44
+aC-COO|[cH0][C](=O)[OH0]|56.0204|45
+aC-OOCH|[CH](=O)[OH0][cH0]|57.028|46
+aC-OOC|[CH0](=O)[OH0][cH0]|56.0202|47
+COO except as above|[CH0](=O)[OH0]|44.009|48
+CH3O|[CH3][OH0]|31.034|49
+CH2O|[CH2;!R][OH0]|30.026|50
+CH-O|[CH;!R][OH0]|29.018|51
+C-O|[CH0;!R][OH0]|28.0102|52
+aC-O|[cH0][OH0]|28.0102|53
+CH2NH2|[CX4H2;!R][NH2]|30.05|54
+CHNH2|[CX4H;!R][NH2]|29.042|55
+CNH2|[CX4H0;!R][NH2]|28.0342|56
+CH3NH|[CH3][NH]|30.05|57
+CH2NH|[CH2][NH]|29.042|58
+CHNH|[CH][NH]|28.034|59
+CH3N|[CH3][NH0]|29.042|60
+CH2N|[CH2;!R][NH0]|28.034|61
+aC-NH2|[cH0][NH2]|28.0342|62
+aC-NH|[cH0][NH]|27.0262|63
+aC-N|[cH0][NH0]|26.0184|64
+NH2 except as above|[NH2]|16.0226|65
+CH=N|[CH;!R]=[NH0]|27.0258|66
+C=N|[CH0;!R]=[NH0]|26.018|67
+CH2CN|[CH2;!R][C]#[N]|40.045|68
+CHCN|[CH;!R][C]#[N]|39.0372|69
+CCN|[CH0;!R][C]#[N]|38.0294|70
+aC-CN|[cH0][C]#[N]|38.0294|71
+CN except as above|[C;!R]#[N]|26.018|72
+CH2NCO|[CH2][NH0]=[C]=[O]|56.0436|73
+CHNCO|[CH][NH0]=[C]=[O]|55.0358|74
+CNCO|[CH0][NH0]=[C]=[O]|54.028|75
+aC-NCO|[cH0][NH0]=[C]=[O]|54.028|76
+CH2NO2|[CH2]N(=O)(=O)|60.032|77
+CHNO2|[CH]N(=O)(=O)|59.024|78
+CNO2|[CH0]N(=O)(=O)|58.016|79
+aC-NO2|[cH0]N(=O)(=O)|58.016|80
+NO2 except as above|N(=O)(=O)|46.005|81
+ONO|[NH](O)=O|30.006|82
+ONO2|[O][NH0](=O)=O|62.004|83
+HCON(CH2)2|[CH2]N([CH2])[CH]=O|71.078|84
+HCONHCH2|[CH2][NH][CH]=O|71.078|85
+CONH2|[CH0](=O)([NH2])|44.0326|86
+CONHCH3|[CH0](=O)([NH][CH3])|58.0592|87
+CONHCH2|[CH0](=O)([NH][CH2])|57.0514|88
+CON(CH3)2|[CH0](=O)([N]([CH3])[CH3])|72.0858|89
+CONCH3CH2|[CH0](=O)([N]([CH3])[CH2])|71.078|90
+CON(CH2)2|[CH0](=O)([N]([CH2])[CH2])|70.0702|91
+CONHCO|[CH0](=O)[NH][CH0](=O)|71.0348|92
+CONCO|[CH0](=O)[NH0][CH0](=O)|70.027|93
+aC-CONH2|[cH0][CH0](=O)[NH2]|56.0436|94
+aC-NH(CO)H|[cH0][NH][CH]=O|56.0436|95
+aC-N(CO)H|[cH0][NH0][CH]=O|55.0358|96
+aC-CONH|[cH0][CH0](=O)[NH]|55.0358|97
+aC-NHCO|[cH0][NH][CH0](=O)|55.0358|98
+aC-(N)CO|[cH0][NH0][CH0](=O)|54.028|99
+NHCONH|[NH][CH0](=O)[NH]|58.0396|100
+NH2CONH|[NH2][CH0](=O)[NH]|59.0474|101
+NH2CON|[NH2][CH0](=O)[NH0]|58.0396|102
+NHCON|[NH][CH0](=O)[NH0]|57.0318|103
+NCON|[NH0][CH0](=O)[NH0]|56.0239|104
+aC-NHCONH2|[NH2][CH0](=O)[NH][cH0]|71.0584|105
+aC-NHCONH|[NH][CH0](=O)[NH][cH0]|70.0506|106
+NHCO except as above|[NH][CH0]=O|43.0248|107
+CH2Cl|[CH2][Cl]|49.48|108
+CHCl|[CH][Cl]|48.472|109
+CCl|[CH0][Cl]|47.464|110
+CHCl2|[CH](Cl)[Cl]|83.925|111
+CCl2|[CH0](Cl)[Cl]|82.917|112
+CCl3|[CH0](Cl)(Cl)[Cl]|118.37|113
+CH2F|[CH2][F]|33.0246|114
+CHF|[CH][F]|32.0168|115
+CF|[CH0][F]|31.009|116
+CHF2|[CH]([F])[F]|51.0148|117
+CF2|[CH0]([F])[F]|50.007|118
+CF3|[CH0]([F])([F])[F]|70.0128|119
+CCl2F|[CH0](Cl)(Cl)[F]|101.915|120
+HCClF|[CH](Cl)[F]|67.4698|121
+CClF2|[CH0](Cl)(F)[F]|85.46|122
+aC-Cl|[cH0][Cl]|47.464|123
+aC-F|[cH0][F]|31.009|124
+aC-I|[cH0][I]|138.911|125
+aC-Br|[cH0][Br]|91.915|126
+-I except as above|[I]|126.9|127
+-Br except as above|[Br]|79.904|128
+-F except as above|[F]|18.998|129
+-Cl except as above|[Cl]|35.453|130
+CHNOH|[CH]=[NH0][OH]|44.0326|131
+CNOH|[CH0]=[NH0][OH]|43.0248|132
+aC-CHNOH|[cH0][CH]=[NH0][OH]|44.0326|133
+OCH2CH2OH|[OH][CH2][CH2][OH0]|61.059|134
+OCHCH2OH|[OH][CH2][CH][OH0]|60.0512|135
+OCH2CHOH|[OH][CH][CH2][OH0]|60.0512|136
+-O-OH|[OH0][OH]|33.0058|137
+CH2SH|[CH2][SH]|47.0994|138
+CHSH|[CH][SH]|46.0916|139
+CSH|[CH0][SH]|45.0838|140
+aC-SH|[cH0][SH]|45.0838|141
+-SH except as above|[SH]|33.0728|142
+CH3S|[CH3][SH0]|47.0994|143
+CH2S|[CH2][SH0]|46.0916|144
+CHS|[CH][SH0]|45.0838|145
+CS|[CH0][SH0]|44.076|146
+aC-S-|[cH0][SH0]|44.076|147
+SO|[SX3H0]=[OH0]|48.064|148
+SO2|[OH0]=[SX4H0]=[OH0]|64.063|149
+SO3 (sulfite)|[OH0]-[SX3H0](=[OH0])-[OH0]|80.062|150
+SO3 (Sulfonate)|[OH0]-[SX4H0](=[OH0])=[OH0]|80.062|151
+SO4 (Sulfate)|[OH0]-[SX4H0](-[OH0])(=[OH0])=[OH0]|96.061|152
+aC-SO|[cH0][SX3H0]=[OH0]|60.075|153
+aC-SO2|[cH0][SX4H0](=[OH0])=[OH0]|76.074|154
+PH (phosphine)|[PX3H]|31.9818|155
+P (Phospine)|[PX3H0]|30.974|156
+PO3 (Phospite)|[PX3H0]([OH0])([OH0])[OH0]|78.971|157
+PHO3 (Phosponate)|[PX4H1](=[OH0])(-[OH0])-[OH0]|79.9788|158
+PO3 (Phosponate)|[PX4H0](=[OH0])(-[OH0])-[OH0]|78.971|159
+PHO4 (Phospate)|[PX4H0](=[OH0])(-[OH1])(-[OH0])-[OH0]|95.9778|160
+PO4 (Phospate)|[PX4H0](=[OH0])(-[OH0])(-[OH0])-[OH0]|94.97|161
+aC-PO4|[cH0][PX4H0](=[OH0])(-[OH0])(-[OH0])-[OH0]|106.981|162
+aC-P|[cH0][PX3H0]|42.985|163
+CO3 (Carbonate)|[CH0](=O)(-[OH0])-[OH0]|60.008|164
+C2H3O|[CH2]([O]1)[CH]1|43.0444|165
+C2H2O|[CH]([O]1)[CH]1|42.0366|166
+C2HO|[CH0]([O]1)[CH]1|41.0288|167
+CH2 (cyclic)|[CX4H2;R]|14.027|168
+CH (cyclic)|[CX4H;R]|13.019|169
+C (cyclic)|[CX4H0;R]|12.011|170
+CH=CH (cyclic)|[CH;R]=[CH;R]|26.038|171
+CH=C (cyclic)|[CH;R]=[CH0;R]|25.03|172
+C=C (cyclic)|[CH0;R]=[CH0;R]|24.022|173
+CH2=C (cyclic)|[CH2]=[CH0;R]|26.038|174
+NH (cyclic)|[NX2H;R]|15.0148|175
+N (cyclic)|[NH0;R]|14.007|176
+CH=N (cyclic)|[CH;R]=[NH0;R]|27.0258|177
+C=N (cyclic)|[CH0;R]=[NH0;R]|26.018|178
+O (cyclic)|[OH0;R]|15.999|179
+CO (cyclic)|[CH0;R]=O|28.01|180
+S (cyclic)|[SX2H0;R]|32.065|181
+SO2 (cyclic)|[SH0;R](=O)(=O)|64.063|182
+>NH|[NX2H;!R]|15.0148|183
+-O-|[OX2H0;!R]|15.999|184
+-S-|[SX2H0;!R]|32.065|185
+>CO|[CX4H0;!R]=O|28.01|186
+PO2|[PX3H0]([OH0])([OH0])|62.972|187
+CH-N|[CH][NH0;!R]|27.0258|188
+SiHO|[SiX4H][OH0]|45.0928|189
+SiO|[SiX4H0][OH0]|44.085|190
+SiH2|[SiX4H2]|30.1016|191
+SiH|[SiX4H]|29.0938|192
+Si|[SiX4H0]|28.086|193
+? (CH3)3N|||194
+N=N (cyclic)|[NH0;R]=[NH0;R]|28.014|195
+Ccyclic=N-|[CH0;R]=[NH0;!R]|26.018|196
+Ccyclic=CH-|[CH0;R]=[CH;!R]|25.0298|197
+Ccyclic=NH|[CH0;R]=[NH]|27.0258|198
+N=O|[NH0]=O|30.006|199
+Ccyclic=C|[CH0;R]=[CH0;!R]|24.022|200
+P=O|[PX5H0]=O|46.973|201
+N=N|[NH0;!R]=[NH0;!R]|28.014|202
+C=NH|[CH0]=[NH]|27.0258|203
+>C=S|[CX3H0]=[SH0]|44.076|204
+aC-CON|[cH0][CH0](=O)[NH0]|54.028|205
+aC=O|[CH0;R]=O|28.01|206
+aN-|[NX3H0;R]|14.007|207
+-Na|[Na]|22.99|208
+-K|[K]|39.098|209
+HCONH|[CH](=O)[NH]|44.0326|210
+CHOCH|[CH]([O]1)[CH]1|42.0366|211
+C2O|[CH0]([O]1)[CH0]1|40.021|212
+SiH3|[SiX4H3]|31.1094|213
+SiH2O|[SiX4H2][OH0]|46.1006|214
+CH=C=CH|[CH]=[CH0]=[CH]|38.0486|215
+CH=C=C|[CH]=[CH0]=[CH0]|37.0408|216
+OP(=S)O|[PX4H0]([OH0])([OH0])=[SH0]|95.037|217
+? R|||218
+CF2cyclic|[CH0;R]([F])[F]|50.007|219
+CFcyclic|[CH0;R]([F])[!F]|31.009|220

ugropy/groupscsv/abdulelah_gani/secondary/secondary.csv

@@ -0,0 +1,130 @@
+(CH3)2CH
+(CH3)3C
+CH(CH3)CH(CH3)
+CH(CH3)C(CH3)2
+C(CH3)2C(CH3)2
+CHn=CHm-CHp=CHk (k,m,n,p in 0..2)
+CH3-CHm=CHn (m,n in 0..2)
+CH2-CHm=CHn (m,n in 0..2)
+CHp-CHm=CHn (m,n in 0..2; p in 0..1)
+CHCHO or CCHO
+CH3COCH2
+CH3COCH or CH3COC
+CHCOOH or CCOOH
+CH3COOCH or CH3COOC
+CO-O-CO
+CHOH
+COH
+CH3COCHnOH (n in 0..2)
+NCCHOH or NCCOH
+OH-CHn-COO (n in 0..2)
+CHm(OH)CHn(OH) (m,n in 0..2)
+CHm(OH)CHn(NHp) (m,n,p in 0..2)
+CHm(NH2)CHn(NH2) (m,n in 0..2)
+CHm(NH)CHn(NH2) (m,n in 1..2)
+H2NCOCHnCHmCONH2 (m,n in 1..2)
+CHm(NHn)-COOH (m,n in 0..2)
+HOOC-CHn-COOH (n in 1..2)
+HOOC-CHn-CHm-COOH (n, m in 1..2)
+HO-CHn-COOH (n in 1..2)
+NH2-CHn-CHm-COOH (n, m in 1..2)
+CH3-O-CHn-COOH (n in 1..2)
+HS-CH-COOH
+HS-CHn-CHm-COOH (n, m in 1..2)
+NC-CHn-CHm-CN (n, m in 1..2)
+OH-CHn-CHm-CN (n, m in 1..2)
+HS-CHn-CHm-SH (n, m in 1..2)
+COO-CHn-CHm-OOC (n, m in 1..2)
+OOC-CHm-CHm-COO (n, m in 1..2)
+NC-CHn-COO (n in 1..2)
+COCHnCOO (n in 1..2)
+CHm-O-CHn=CHp (m,n,p in 0..3)
+CHm=CHn-F (m,n in 0..2)
+CHm=CHn-Br (m,n in 0..2)
+CHm=CHn-I (m,n in 0..2)
+CHm=CHn-Cl (m,n in 0..2)
+CHm=CHn-CN (m,n in 0..2)
+CHn=CHm-COO-CHp (m,n,p in 0..3)
+CHm=CHn-CHO (m,n in 0..2)
+CHm=CHn-COOH (m,n in 0..2)
+aC-CHn-X (n in 1..2) X: Halogen
+aC-CHn-NHm (n in 1..2; m in 0..2))
+aC-CHn-O- (n in 1..2)
+aC-CHn-OH (n in 1..2)
+aC-CHn-CN (n in 1..2)
+aC-CHn-CHO (n in 1..2)
+aC-CHn-SH (n in 1..2)
+aC-CHn-COOH (n in 1..2)
+aC-CHn-CO- (n in 1..2)
+aC-CHn-S- (n in 1..2)
+aC-CHn-OOC-H (n in 1..2)
+aC-CHm-NO2 (n in 1..2)
+aC-CHn-CONH2 (n in 1..2)
+aC-CHn-OOC (n in 1..2)
+aC-CHn-COO (n in 1..2)
+aC-SO2-OH
+aC-CH(CH3)2
+aC-C(CH3)3
+aC-CF3
+(CHn=C)(cyc)-CHO (n in 0..2)
+(CHn=C)cyc-COO-CHm (n,m in 0..3)
+(CHn=C)cyc-CO- (n in 0..2)
+(CHn=C)cyc-CH3 (n in 0..2)
+(CHn=C)cyc-CH2 (n in 0..2)
+(CHn=C)cyc-CN (n in 0..2)
+(CHn=C)cyc-Cl (n in 0..2)
+CHcyc-CH3
+CHcyc-CH2
+CHcyc-CH
+CHcyc-C
+CHcyc-CH=CHn (n in 1..2)
+CHcyc-C=CHn (n in 1..2)
+CHcyc-Cl
+CHcyc-F
+CHcyc-OH
+CHcyc-NH2
+CHcyc-NH-CHn (n in 0..3)
+CHcyc-N-CHn (n in 0..3)
+CHcyc-SH
+CHcyc-CN
+CHcyc-COOH
+CHcyc-CO
+CHcyc-NO2
+CHcyc-S-
+CHcyc-CHO
+CHcyc-O-
+CHcyc-OOCH
+CHcyc-COO
+CHcyc-OOC
+Ccyc-CH3
+Ccyc-CH2
+Ccyc-OH
+>Ncyc-CH3
+>Ncyc-CH2
+AROMRINGs1s2
+AROMRINGs1s3
+AROMRINGs1s4
+AROMRINGs1s2s3
+AROMRINGs1s2s4
+AROMRINGs1s3s5
+AROMRINGs1s2s3s4
+AROMRINGs1s2s3s5
+AROMRINGs1s2s4s5
+PYRIDINEs2
+PYRIDINEs3
+PYRIDINEs4
+PYRIDINEs2s3
+PYRIDINEs2s4
+PYRIDINEs2s5
+PYRIDINEs2s6
+PYRIDINEs3s4
+PYRIDINEs3s5
+PYRIDINEs2s3s6
+(CHn=CHm)cyc-COOH
+AROMRINGs1s2s3s4s5
+aC-NHCOCH2N
+(N=C)cyc-CH3
+aC-CONH(CH2)2N
+aC-SO2NHn (n>=0;n<3)
+aC-SO2NHn (n>=0;n<3)
+aC-SO2NHn (n>=0;n<3)